Spontaneous ignition of gas turbine lubricants at temperatures below their standard auto-ignition temperatures

Prepared by the Health and Safety Executive

RR1076 Research Report

© Crown copyright 2016

Prepared 2014 First published 2016

You may reuse this information (not including logos) free of charge in any format or medium, under the terms of the Open Government Licence. To view the licence visit www.nationalarchives.gov.uk/doc/open-government-licence/, write to the Information Policy Team, The National Archives, Kew, London TW9 4DU, or email psi@nationalarchives.gsi.gov.uk.

Some images and illustrations may not be owned by the Crown so cannot be reproduced without permission of the copyright owner. Enquiries should be sent to copyright@hse.gsi.gov.uk.

This report and the work it describes were funded by the Health and Safety Executive (HSE). Its contents, including any opinions and/or conclusions expressed, are those of the authors alone and do not necessarily reflect HSE policy.

There have been a number of incidents resulting in lubricating oil leaking in offshore gas turbine enclosures which could ignite if they came into contact with hot surfaces below their Auto Ignition Temperature (AIT). To assess the risk of auto-ignition, standard minimum AITs are used. However, AITs under industrial conditions are difficult to calculate and can be less than these standard values.

This report describes research using a Spontaneous Combustion Calorimeter developed to study spontaneous ignition. Preliminary tests were done for a range of process conditions that can influence minimum AITs for a number of gas turbine lubricating oils. These showed that ignition can occur at temperatures well below the standard minimum AIT. This indicates that if manufacturers rely on standard AITs at the design stage of gas turbines and enclosures, it may lead to a system that is likely to increase the ignition probability of any flammable release. To confirm these findings, further tests would be needed over a wider temperature range and under conditions which more closely represent the conditions in gas turbine enclosures.

Until AITs under industrial conditions are understood and addressed in design criteria, dutyholders will need to err on the side of caution in identifying and adequately controlling potential ignition sources.

HSE Books

Lorem ipsum dolor sit amet consectetuer adipiscing elit

Dr TJ Snee, R Braddock and Dr JT Allen Health and Safety Executive Harpur Hill Buxton Derbyshire SK17 9JN

Spontaneous ignition of gas turbine lubricants at temperatures below their standard auto-ignition temperatures

1

ACKNOWLEDGEMENTS

Acknowledgement is given by Dr T Snee, primary author, and Dr J T Allen of the extensive

practical and other assistance provided by Mr R Braddock throughout the course of this work.

This includes production of the design and build specification documentation contained within

this report as Appendix A.

2

EXECUTIVE SUMMARY

In order to determine how process conditions can influence minimum auto-ignition

temperatures in gas turbine enclosures, and other industrial installations, a novel type

of calorimeter has been developed at HSE's Buxton research laboratory. This instrument,

a Spontaneous Combustion Calorimeter (SCC), has been used to investigate the

ignition properties of pure substances, such as n-heptane and compare the results with minimum

ignition temperatures determined using standard methods. This was followed by a

more detailed investigation of the ignition of gas-turbine lubricants.

Standard minimum auto-ignition temperatures (A.I.T.s) are included in manufacturers’ safety

data sheets and are widely used to assess the risk of auto-ignition in industrial installations.

Minimum ignition temperatures under industrial conditions are less than the standard A.I.T.s

because increases in scale reduce the rate of heat loss per unit volume. It is difficult to calculate

the reduction in minimum ignition temperature associated with increasing scale. Many risk

assessment apply an arbitrary margin of safety between process temperature and the standard

A.I.T.

Results of experiments on n-heptane were consistent with previously published data and

confirmed that the SCC can be used to develop kinetic models for the slow oxidation processes

which lead to ignition.

The gas turbine lubricants were found to auto-ignite at relatively high temperatures > 350oC,

with no evidence of slow oxidation preceding ignition. The results indicated that aging had no

strong influence on the minimum ignition temperature of the lubricants.

It is generally assumed that the standard method of measuring A.I.T. yields values which are

lower than those obtained from alternative laboratory-scale procedures. The minimum ignition

temperature when fuel is added to a static, enclosed, volume of heated air in the standard

apparatus is lower than that obtained at a heated surface or when the fuel-air mixture flows at a

significant rate through the test vessel. However, results for the gas turbine lubricants in the

SCC indicate that, if cold air is added slowly to heated fuel (rather than fuel added to hot air in

the standard procedure), ignition can occur at temperatures well below the standard A.I.T. This

would imply that operators/Dutyholders may still have a significant fire risk even where they

have, in good faith, employed all suitable engineering and other controls to reduce the

theoretical risk to ALARP levels, based on published A.I.T.s.

The experiments on the gas turbine lubricants need to be repeated over a wider temperature

range and under conditions which more closely represent the conditions in the gas turbine

enclosures. If further experiments support the preliminary interpretation of the initial

experiments, the wider implications for other types of industrial installation should be

examined.

3

4

CONTENTS PAGE

1. INTRODUCTION .................................................................... 5

2. LABORATORY-SCALE APPARATUS .................................. 6

2.1 Standard Auto-ignition Temperature (A.I.T.) 6 2.2 Spontaneous Combustion Calorimeter (SCC) 6

3. N-HEPTANE ........................................................................... 8

3.1 Experimental procedure 8 3.2 Results 8

4. GAS TURBINE LUBRICANTS ............................................. 11

4.1 Experimental procedure. 11

4.2 Experimental results 11 4.3 Comparison with standard A.I.T. 14 4.4 Temperature dependence of Peak B 15

5. DISCUSSION........................................................................ 17

6. CONCLUSIONS AND IMPLEMENTATION .......................... 18

7. REFERENCES ..................................................................... 19

APPENDIX A - HSL SCC .............................................................. 20

A1 – Calorimeter Design and Build 20 A2 - Calorimeter control system 39

A3 - Crydon D2450 Solid state relay specification data sheet 43

5

1. INTRODUCTION

The work described in this report forms part of an investigation of combustion hazards in gas turbine

enclosures requested by HSE Energy Division (ED3.2) in response to a series of fires that have

occurred at offshore installations. Other work in this investigation (Fletcher J, 2014) has concerned

itself with identifying common causes or modes of failure, based on reports of previous fires, and

reviewing relevant advice available to turbine Dutyholders.

The objectives of the study covered in this report, which concern auto-ignition hazards, are:

a) the design and construction of a laboratory-scale apparatus, and

b) the development of an associated methodology for assessing the likelihood and consequences

of auto-ignition in gas turbines.

The reliability of the methodology was to be determined using substances with well-established

physical and chemical properties. A calorimeter, designed for the investigation of auto-ignition, has

been constructed at the HSE's Buxton research laboratory. After a review of the initial experimental

data on n-heptane, Energy Division (ED 3.2) requested further work on samples of used

turbine oil, obtained from a number of offshore installations, and a fresh sample of turbine

oil from the manufacturer.

Process temperatures which can result in auto-ignition are determined by the operating environment

and the physical and chemical properties of the process fluids. The heat transfer characteristics of the

process environment or hot surface can have a strong influence on the minimum ignition temperature.

In general, an increase in vessel volume will result in a reduction in auto-ignition temperature.

Ignition at hot surfaces requires a higher temperature than that necessary to produce ignition in a

closed vessel.

Minimum auto-ignition temperatures (A.I.T), measured in laboratory-scale equipment, are used both

by manufacturers to minimise the initial risk of ignition through the design process and also by

operators/Dutyholders to assess the residual risk of ignition in industrial processes. The standard

apparatus for measuring A.I.T. is considered in the first section of this report, followed by a

brief description of the new Spontaneous Combustion Calorimeter (SCC) developed at HSE's Buxton research laboratory. Detailed SCC design and build information is presented in Appendix A.

The performance of the SCC has been assessed using n-heptane. Results of these experiments are

reported and compared with data from an earlier version of the SCC. This discussion includes a set of

curves which illustrate how kinetic data can be obtained from the SCC and used to calculate minimum

process temperature that can lead to ignition.

The experimental procedure and results for the gas turbine lubricants are described in the main section

of this report. The results demonstrate how reliance by manufacturers on the standard A.I.T.s at the

design stage of gas turbines and their enclosures can lead to a system that is likely to increase the

ignition probability of any flammable release. Effectively, operators/Dutyholders may still have a

significant fire risk even where they have, in good faith, employed all suitable engineering and other

controls to reduce the theoretical risk to ALARP levels.

6

2. LABORATORY-SCALE APPARATUS

2.1 STANDARD AUTO-IGNITION TEMPERATURE (A.I.T.) In the standard methods for determining A.I.T., a small quantity of fuel is injected into a uniformly-

heated flask held in a thermostatic oven. Fuel is injected, over a range of oven temperatures, until

ignition is detected by the observation of a flame or a sharp increase in temperature inside the flask.

Measurements are repeated, with various amounts of fuel, in order to determine the minimum oven

temperature necessary for ignition. The standard methods use a conical or spherical flask ranging in

size from 125 to 250 ml.

The standard A.I.T. cannot be used directly to determine a safe equipment operating temperature

because of the effect of scale and the differences between process conditions and the conditions in the

standard apparatus.

Differences can arise due to, for example;

variations in the degree of agitation

non- uniform temperature and concentration distributions in the process vessel

effects of flow on the rate of oxidation of the fuel air mixture

variation in the geometry of the process vessel

reduced rates of oxidation in the presence of inert gas

In order to quantify these effects it is necessary to obtain experimental data under conditions which

can be related to the conditions in the industrial process. The standard measurement of A.I.T. needs to

be interpreted using measurements of the temperature and concentration dependence of the slow

oxidation reactions which precede ignition. The SCC has been designed to provide detailed data

which can be used to develop a kinetic model for the oxidation reactions. However, in many cases, it

is sufficient to use the instrument to investigate how variations in specific process parameters can

affect the conditions for ignition. In this way, it may be possible to determine the margin of safety

which needs to be established between the process temperature and the standard A.I.T.

2.2 SPONTANEOUS COMBUSTION CALORIMETER (SCC) The SSC is shown schematically in Figure 1. The sample is held in a thin-walled stainless-steel test

cell which is surrounded by guard heaters mounted around a copper tube. Adiabatic conditions are

achieved by matching the temperature of the guard heaters to that of the sample. The calorimeter can

also operate in isoperibolic mode, i.e. where the surroundings of the reaction mass are maintained at a

constant temperature so that any exothermic or endothermic changes produce a corresponding

temperature change in the reactor. Such operation is achieved here by setting the guard heat to

maintain a constant wall temperature. Peripheral equipment is provided for various modes of sample

injection and adjustment of the pressure and composition of gas and vapour in the test cell. The

control systems and detailed technical specification for the instrument is described in Appendix A.

7

Figure 1 - Schematic diagram of the Spontaneous Combustion Calorimeter.

Further detail is provided in Appendix A.

T

T

Air Vacuum

P

Copper Calorimeter Chamber

Hotplate Stirrer

Exhaust

Test Cell

T

T

T

T

8

3. N-HEPTANE

In order to evaluate the performance of the SCC, experiments were carried out on n-heptane. This

pure substance has a flash point of -4

oC and a standard A.I.T of 223

oC.

3.1 EXPERIMENTAL PROCEDURE The experiments were performed by first setting the instrument to maintain a constant wall

temperature (isoperibolic conditions) and allowing sufficient time for the sample-can to reach the set

point. The sample-can was then connected to a vacuum pump and a partial vacuum was maintained in

order to remove volatile contaminants. The can was then isolated from the pump and connected to a

small reservoir which was open to atmosphere and contained n-heptane. The sample was drawn into

the can, followed by sufficient air to restore atmospheric pressure and then the connection to the

reservoir was closed.

3.2 RESULTS Figure 2 and Figure 3 show the results of experiments at set points of 230 and 249

oC, respectively.

After sample injection, both experiments show a gradual increase in temperature with the pressure

remaining at the initial value of 1 Bara. This was followed by a sharp increase in pressure and

temperature, indicating that ignition had occurred. After ignition, the temperature returned gradually

to the set point, but the pressure remained above atmospheric. This excess pressure is attributable to

gaseous products of combustion.

Results from the two experiments are summarised in Table 1. The data must be interpreted carefully

because of the limitations in response time of the pressure and temperature transducers, relative to the

timescale over which processes are occurring, at the time of ignition. The more gradual changes in

temperature, during the induction period, can be monitored accurately and these changes provide an

indication of the rate of acceleration of the slow oxidation reactions that lead to ignition. Subject to

the outcome of this work, higher response rate transducers may wish to be considered.

Observation of ignition at the set point temperatures of 230 and 249oC is consistent with experimental

data obtained previously, under adiabatic conditions (Snee & Montserrat, 2010). Under isoperibolic

and adiabatic conditions, temperature increases are lower than would be observed in an isolated

system. This is due to the thermal dilution due to heat transfer to the sample-can. Thermal dilution

allows measurement of rates of slow combustion over a wider temperature range. The temperature

increase in an isolated system can be calculated using the adiabatic data along with calculated values

for the thermal capacity of the fuel-air mixture and the thermal capacity of the sample-can.

Table 1 - Experimental results for n-heptane.

Set point 230 249

Initial temperature (oC) 231.3 249.0

Maximum temperature (oC) 245.1 267.3

Temperature increase (oC) 13.8 19.5

Initial pressure (bara) 0.995 0.995

Maximum pressure (bara) 1.291 1.233

Pressure increase (bara) 0.296 0.238

Ignition delay (sec) 28.3 s 16.0 s

9

220

230

240

250

260

270

242000 243000 244000 2450000.0

0.5

1.0

1.5

200 micro litres heptane

isoperibolic: 230°C

Pres

sure

(bar

a)

Time (100 ms)

pressure

Tem

pera

ture

(°C)

sample temperature wall temperature

28.3 sec

sampleinjection

Figure 2 - Temperature and pressure changes during ignition of n-heptane in the SCC at 230oC.

240

250

260

270

53000 54000 55000 560000.0

0.5

1.0

1.5

200 micro litres heptane

isoperibolic: 249°C

Pres

sure

(bar

a)

Time (100 ms)

pressure T

empe

ratu

re (°

C)

sample temperature wall temperature

16 sec

sampleinjection

Figure 3 - Temperature and pressure changes during ignition of n-heptane in the SCC at 249oC.

Kinetic parameters for slow combustion reactions can be obtained from a series of experiments under

adiabatic conditions over a range of initial temperatures. These measurements have been performed

using an earlier version of the SSC as part of the previous project. Figure 4 shows results from a series

of experiments with heptane-air mixtures over the temperature range from 190 to 230oC. The results

0 10 20 30 40170

180

190

200

210

220

230

240

1.0

1.5

2.0

2.5

3.0

0 20 40

180

210

240

ignition

Tem

pera

ture

(°C

)

Initial temperature 190°C 200°C 210°C 220°C 230°C

Time (min.)

Fitted curves for autocatalysis

Pre

ssur

e (b

ara)

corresponding pressure records

10

in Figure 4 demonstrate how progressive increases in the initial temperature and the corresponding

rates of slow oxidation eventually lead to supercritical conditions and ignition. The temperature

profile during ignition is distinct from the s-shaped temperature-time curves observed under

subcritical conditions. Figure 4 shows that there was no substantial pressure increase during slow

oxidation. The rapid pressure increase (to 1.4 Bar) was only observed during ignition.

Figure 4 - Adiabatic slow combustion and ignition data for n-heptane. The fitted curves, plotted

in black, are derived from a kinetic model of the slow combustion processes leading to ignition.

Gas turbine fuels are likely to auto-ignite over a temperature range similar to the range over which

slow rates of oxidation of n-hexane can lead to ignition. It would be useful to obtain adiabatic data on

these fuels so that the potential hazards due to fuel leaks in the turbine enclosure could be assessed.

Kinetic analysis of the results for n-heptane in Figure 4 can be used to calculate the minimum process

temperature that can lead to ignition in a system. Alternatively, the results could also be used in

C.F.D. modelling of temperature and concentration distributions in industrial installations, although

these are beyond the scope of the current work. Regulatory authorities and industrial consortia may

utilise this type of analysis in order to raise awareness of the problems that can arise if a standard

A.I.T. is misinterpreted at the design criteria stage, and subsequently. Results for gas turbine

lubricants in the following section demonstrate how important findings can be obtained without full

kinetic analysis.

0 10 20 30 40170

180

190

200

210

220

230

240

1.0

1.5

2.0

2.5

3.0

0 20 40

180

210

240

ignition

Tem

pera

ture

(°C

)

Initial temperature 190°C 200°C 210°C 220°C 230°C

Time (min.)

Fitted curves for autocatalysis

Pre

ssur

e (b

ara)

corresponding pressure records

11

4. GAS TURBINE LUBRICANTS

Energy Division (ED 3.2) provided 3 samples of lubrication oil from working offshore platform

compressor turbines. The bottles were labelled C1, C2 and C3. The bottles contained used oil but it

has not been possible to establish the exact provenance of the samples or the duration over which the

oil samples would have been subject to elevated temperatures in the turbine enclosures.

Further, in order to allow comparative aging tests to be undertaken, HSE obtained a sample

of representative unused lubricating oil, AeroShell Turbine Oil 500, direct from the manufacturer.

The flash point of each sample was measured using a Seta Multiflash apparatus and following BS EN

ISO 3679-2004. The standard A.I.T. of one of the samples (C1) was measured following the

procedure set out in DIN51 794: Testing of Mineral Oil Hydrocarbons. Results of these

measurements are summarised in Table 2 along with additional data obtained from the M.S.D.S for

the new oil.

Table 2 - Summary of test results on oil Sample C1 when tested in accordance with BS EN ISO

3679-2004.

C1 C2 C3 AeroShell

Flash point (oC) 209.5 209.5 209.5 260

(typical)

AIT (oC

) 340 >320

Density (kg m3) 1.005

Vapour density (air =1) >1

(estimated)

Initial boiling point (oC) >280

(estimated)

Upper/Lower Flammability or

Explosion Limits (oC

)1 – 10 %

(V)

4.1 EXPERIMENTAL PROCEDURE The gas-turbine lubricants were tested using a procedure similar to that followed for the experiments

on n-heptane. Fuel concentrations and temperature set-points were chosen over ranges close to the

values at which ignition was observed in the standard A.I.T. apparatus. Initial experiments showed a

significant temperature increase when air was used to flush the sample can at the end of the test.

Further experiments were performed using controlled flow-rates of air or nitrogen to flush the can in

order to investigate the mechanism that may be responsible for these increases in temperature.

4.2 EXPERIMENTAL RESULTS

4.2.1 Used oil Figure 5 shows the results of an experiment in which 0.3 ml of lubricant (sample C3) was drawn into

the test can when the wall temperature had stabilised at 370oC. Immediately after sample injection, the

temperature increased rapidly to 390oC (Peak A) and then gradually returned to the initial value of

370oC. After a further 15 min., a slow flow of air was established through the test can with a pressure

of 1.25 Bara at the inlet. This resulted in a rapid increase in sample temperature to 410oC (Peak B).

The shape of this peak was similar to the peak associated with fuel injection.

12

The pressure trace in Figure 5 shows an initial increase from 0.63 Bara to 1 Bara as fuel and air are

drawn into the sample can after partial evacuation. The increase in pressure to 1.25 Bara, at the end of

the experiment, is due to the application of air pressure to give a controlled rate of flushing.

Figure 5 - Chart to show temperature and pressure measured by the SCC during tests on

sample C3.

Experiments were performed under the following conditions

Set point temperature: 350oC, 360

oC and 370

oC

Fuel injected: 0.1ml, 0.2ml, 0.3ml and 0.4ml

Wait time between fuel injection

and air injection: 5min, 10min. and 15min.

The experiments were designed to determine the significance of a range of factors which could affect

thermal stability. Results and tentative conclusions (in italics) are listed below:

1. Temperature

Two peaks (A and B) were observed for all experiments with a set point of 370oC

At 360oC, Peak B was observed for all experiments, but Peak A was absent, apart from small

temperature changes observed for experiments using 0.3 and 0.4ml of fuel

At 350oC only Peak B was observed.

The minimum ignition temperature when used oil was added to air in the SSC was 360oC

2. Wait time

In experiments in which only the wait time between fuel and air injection was varied, there

were no significant changes in the temperature increase associated with Peak B

The wait time had no significant effect on the thermal stability of the mixture in the sample can.

0

1

2

3

4

5

800 1000 1200 1400 1600 1800350

360

370

380

390

400

410

420

C3 0.3 ml 15 min waitTe

mpe

ratu

re (°

C)

Time (sec.)

temperature

Pre

ssur

e (b

ara)

pressure

fuelinjection

air injection

Peak A Peak B

wait time

13

3. Nitrogen injection (Shown in Figure 6)

Only a small temperature change was observed when nitrogen was injected, instead of air, at

the end of the wait time.

Peak B was associated with oxidative self-heating.

4. Source of sample

At the same set point, similar results were obtained for Samples C1, C2 and C3.

Aging of the lubricant had no strong influence on the thermal stability

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

4600 4700 4800 4900 5000 5100 5200 5300330

360

390

420

0 .2 ml Nitrogen Flush

Tem

pera

ture

(°C

)

Time (s)

temperature

nitrogeninjection

pressure

Pre

ssur

e (b

ara) fuel

injection

Figure 6 - Temperature and pressure records for an experiment using 0.2ml of Sample C1

where nitrogen, instead of air, was injected at the end of the test.

14

2

4

6

10 12 14 16 18 20 22350

360

370

380

390

400

410

420

Fresh LubricantTr

empe

ratu

re (

o C)

Time (min.)

temperature

Peak APeak B

fuelinjection

airinjection

Pre

ssur

e (b

ara)

pressure

Figure 7 - SCC data for the ignition of AeroShell 500 at an initial temperature of 360oC.

4.2.2 Fresh Sample Results for AeroShell 500 at an initial temperature of 360

oC are shown in Figure 7. The temperature

increase associated with Peak B in this experiment was higher than that observed with used samples at

360oC. This indicates that the minimum ignition temperature of the fresh lubricant in the SCC may be

less than the value of 360oC observed for the used samples. Further experiments are needed in order

to confirm this conclusion. The relationship between ignition in the SCC and values obtained from the

standard A.I.T. apparatus is discussed in the following section.

4.3 COMPARISON WITH STANDARD A.I.T. Fuel is injected into hot air in order to determine minimum ignition temperature in the standard

apparatus for measuring A.I.T. A similar procedure was followed using the SCC but improved mixing

is obtained by partial evacuation of the test can prior to sample injection. Partial evacuation also gives

better control of the ratio of fuel to air. Auto-ignition of Sample C1 in the standard apparatus occurred

at a minimum temperature of 340oC. With a smaller test cell in the SCC, the minimum temperature

necessary to produce ignition (associated with Peak A) was 360oC. The increased surface to volume

ratio of the SCC sample can, compared with that of the 200 ml conical flask in the standard apparatus,

is consistent with the observed increase in minimum ignition temperature. Published data on the

variation of A.I.T. with experimental conditions indicate a linear correlation between minimum

ignition temperature and the surface to volume ratio of the test vessel (Swarts D E, Orchin M, 1956).

Application of this correlation, using the surface to volume ratio of the SCC test can, indicates that the

minimum ignition temperature in the SCC will be 21.1oC higher than that observed in the standard

apparatus. This is consistent with the experimentally observed difference of 20 o

C between minimum

ignition temperatures measured in the SCC and the standard apparatus.

The flow of air used to flush the test can in the SCC was much less than that which would be used to

flush the conical flask in the standard apparatus. At the end of a test in the standard apparatus, rapid

flushing quickly reduces the temperature of the contents of the flask and no second peak is observed.

Experiments using a range of air-inlet pressures would provide an indication of the range of air flow-

rates which can lead to ignition. If the air flow through a gas turbine is interrupted at elevated

temperature, gradual resumption of ventilation could be hazardous.

15

4.4 TEMPERATURE DEPENDENCE OF PEAK B The results for the used lubricants and the fresh sample of AeroShell 500 indicate that ignition can be

induced by injecting air into heated fuel as well as by fuel injection into hot air. Further work is

needed to confirm this conclusion and establish whether this behaviour is restricted to gas turbine

lubricants.

62 64 66 68 70 72 74 76 78 80 82 84 86 88 90310

320

330

340

350

360

370

380

390

400

410

0.2 ml 10 min wait

Tem

pera

ture

(oC

)

Time (min.)

Initial temperature

350oC340oC330oC315oC360oC370oC

Peak A

Figure 8 - Temperature records from a series of experiments in the SCC when the initial

temperature was progressively reduced from 370oC to 315

oC. (Between each data set the time

axis has been shifted by 1 minute, in order to facilitate comparison of the results.)

Figure 8 shows the results for Sample C1 over a range of initial temperatures. The experiments with

initial temperatures from 315oC to 350

oC were performed with a sample volume of 0.2 ml. The

experiments at 360 o

C and 370oC had sample volumes of 0.4ml and 0.3ml, respectively. Peak A was

observed in these two experiments and this is associated with auto ignition when fuel is added to hot

air. Peak A diminishes significantly between 370oC and 360

oC and has not been observed at

temperatures below 360oC. By contrast, Peak B, which is associated with the addition of cold air to

the heated contents of the sample can, is observed at initial temperatures as low as 315oC. At 315

oC,

the temperature increase associated with Peak B is relatively large (29.7 o

C) and indicates that the

peak will still be observed at initial temperatures substantially below 315oC.

The chemical composition of the gas turbine lubricants has not been investigated. The data sheet for

AeroShell 500 indicates that the oils comprise a blend of synthetic esters and additives. Only

“estimated” and “typical values” are provided for the vapour density and flammability limits but the

data indicates minimum ignition temperatures in the SCC were obtained under very fuel rich

conditions. It is possible that slow addition of air may reduce the minimum ignition temperature but a

review of published data has produced no evidence to support this hypothesis. Under fuel rich

conditions, at elevated temperature, air slowly added to the hot fuel may react and generate heat at a

rate sufficient to exceed any heat losses (‘cooling’) caused by the introduction of the cold air, thus

resulting in a net increase in temperature (heat). This increase in temperature would increase the rate

of oxidation, which may ultimately lead to ignition. This would need to be studied further

theoretically and experimentally, but is beyond the scope of the current project.

16

Experiments using a range of air-inlet pressures, during flushing, would provide insights into the

physical and chemical processes associated with Peak B. Increasing air flow rates will ultimately

cause cooling and reduce the likelihood of ignition. Further experiments should be performed to

confirm that the rapid rates of temperature rise associated with Peaks A and B are due to flame

propagation.

Detailed evaluation of the experimental results in relation to the temperature distribution and pattern

of air flow in gas turbine enclosures is beyond the remit for the present investigation.

17

5. DISCUSSION

The mechanism associated with the oxidation of hydrocarbons varies according to their chemical

structure. Hydrocarbons with a carbon-chain length greater than 5 have relatively low A.I.T.s,

typically <300OC and with significant delay between sample injection and ignition. The results for n-

heptane have shown how slow combustion can be observed at temperatures well below the A.I.T.

For substances with A.I.T.s greater than 300oC, the delay between sample injection and ignition tends

to be relatively short, with no evidence of slow combustion preceding ignition. The results for gas

turbine lubricants display this type of behaviour. In the SSC the sharp increases in temperature

associated with Peaks A and B occur immediately after fuel and air injection, respectively. There is no

evidence of slow combustion, preceding the peaks, or at temperatures below the minimum ignition

temperature. Adiabatic operation, at these temperatures, would not provide useful kinetic data. In this

temperature regime, controlled flow of the fuel air mixtures through a heated tube can be used to

observe the rate of vapour phase oxidation. The SCC can be configured to provide this type of

measurement.

It must be emphasized that this discussion is based on a tentative interpretation of the origin of Peak

B. The results have been obtained from new equipment only recently commissioned. Repeat experiments must be performed in order verify or dismiss the current hypothesis.

18

6. CONCLUSIONS AND IMPLEMENTATION

6.1 A new type calorimeter for studying spontaneous ignition has been constructed at HSE's Buxton

research laboratory has been shown to produce reliable data on slow oxidation and ignition of n-heptane/air

mixtures.

6.2 The calorimeter has been used successfully to observe the effects of a range of process

conditions which can influence minimum ignition temperatures.

6.3 A preliminary investigation of ignition properties of gas turbine lubricants has

identified measurements and precautions which could reduce the likelihood of ignition in the turbine

enclosures.

6.4 The experimental evidence indicates that, in laboratory-scale equipment, ignition can occur at

temperatures well below the minimum ignition temperature in the standard apparatus for

measuring AIT.

6.5 Reliance by manufacturers on the standard A.I.T.s at the design stage of gas turbines and

their enclosures can lead to a system that is likely to increase the ignition probability of any

flammable release. Effectively, operators/Dutyholders may still have a significant fire risk even

where they have, in good faith, employed all suitable engineering and other controls to reduce the

theoretical risk to ALARP levels. This may go some way to explaining the continuing significant

levels of such events (Fletcher J, 2014), despite the introduction of recent guidance.

6.6 It is recommended that in order to minimise the likelihood of

ignition, manufacturers/operators/Dutyholders should ensure the adequate design, operation,

modification and maintenance of oil, lubrication and fuel systems’ integrity, which could

otherwise result in a flammable release and/or source of ignition.

6.7 Assessments of residual ignition risk, based on the standard A.I.T, should be checked

to determine whether the process conditions are similar to those which produced reduced

ignition temperatures on laboratory scale.

6.8 In pragmatic terms, until any ramifications of this work are confirmed and addressed

in revised design criteria, Dutyholders will need to better identify and adequately control

potential ignition sources. Specific areas may include exhaust manifolds and pipework, turbines

and their enclosures, and heating elements. In practice, however, a fire risk may arise wherever an

accidental release (liquid, vapour or aerosol) of oil, lubricants or fuel is able to impinge on a

sufficiently hot surface. Based on the work in this report, further studies may redefine what

temperatures qualify as ‘sufficiently hot’ but current standard A.I.T.s should not automatically be

used as a ‘safe’ guideline.

19

7. REFERENCES

Fletcher J (2014). An Examination of Incidents Involving Gas Turbines and the Guidance That

Applies to Them, HSL Report FP/09/11, 2014.

Snee T J and Montserrat Siscart J RJ (2008). Assessment of the Critical Conditions for Slow

Oxidation and Autoignition in Large Process Vessels.13th International Symposium on Loss

Prevention in the Process Industries, Brugge, 2010.

Swarts D E, Orchin M (1956). Vapour-Phase Oxidation and Spontaneous Ignition –

3579Correlation and Effect of Variables. U.S. National Advisory Committee for Aeronautics.

Technical Note 3579, 1956.

20

APPENDIX A - HSL SCC

A1 – CALORIMETER DESIGN AND BUILD The Offshore Gas Turbine auto-ignition project required a bespoke adiabatic calorimeter to be

built by HSE. This appendix details the design and build of that calorimeter – a Spontaneous

Combustion Calorimeter (SCC). It is the aim of this overall project to gain an understanding of the

potential for ignition of flammable oil mists formed by normal operation of offshore gas turbines

and aged turbine lubricant oils in hot environments. The SCC would allow both investigation of auto-

ignition processes and comparative testing of specific oil samples acquired from dutyholders who

operate offshore gas turbines, where these oils are used for turbine lubrication.

The calorimeter, using IKA Labworldsoft software, was designed to run in either isothermal mode or

adiabatic mode.

Figure A1.1 shows a high level block diagram of the key elements of the calorimeter.

Figure A1.1 - Block diagram to show the flow of information between key parts of the

calorimeter.

Figure A1.2 shows the calorimeter being prepared for a test run, with the top plate assembly, sample

reservoir and vacuum pump assembly removed from the main body of the calorimeter.

Figure A1.2 - Photograph to show the calorimeter in-situ, with top plate assembly removed.

Calorimeter

chamber/body

Solid state relay control

box & temperature

transmitters

Control & data

acquisition

software

Video recording

& monitoring

21

When fully assembled, the calorimeter body was secured in place between a top and a bottom

stainless plate using threaded bar and appropriate nuts. A pressure tight seal was achieved using

suitable gasket materials at the joints. The copper body of the calorimeter and heaters were clad in

glass fibre wool insulation to minimise heat loss from the heated chamber (Figure A1.3). Early trial

runs highlighted the importance of routing the electrical cables and thermocouple extensions to the

outside of the insulation in order to reduce insulation burning damage. In Figure A1.3 therefore, the

cabling was routed away from the metal work, and the small signal and mains voltage were separated.

Figure A1.3 - Photograph to show the calorimeter hardware fully assembled.

The main body of the calorimeter was earthed by clamping an earth tag to the bottom steel plate, to

which all metalwork was connected by the threaded bar. The earthing tag was connected to mains

earth via the solid state relay box.

A1.1 - Electrical connection After each test, all mains wiring was visually checked for signs of heat induced damage. Initial test

runs showed that vulnerable areas of wiring were the glass braided single core cables connecting the

heater bands, and their crimped ring connectors. Thermocouple operation was tested after each run,

in situ, using a calibrated handheld thermometer, and this was then verified by the data logger PC.

22

Figure A1.4 - Photograph of the unfused mains plug socket used to change between glass braided cable and PVC insulated mains power cable.

The calorimeter’s band heaters received switched mains power from the solid state relay / control box

via twin core insulated cable. The change over from normally insulated mains cable to 2 x single

cored glass insulated cable was achieved using an un-fused mains plug (Figure A1.4).

Glass insulated cable

PVC insulated cable from

Solid State Relay box

23

A1.2 - Heater Power Consumption For the 3 x band heaters and the 1 x top plate ring heater, the maximum electrical power consumption

(assuming all four heaters are powered for 100% of the PWM duty cycle) was calculated as follows

based on:

Band heater, where:

Max power output = 770 W

Mains voltage = 230 V AC

Ring heater, where:

Max power output = 550 W

Mains voltage = 230 V AC

Four heaters (three band heaters and one ring heater) were powered in parallel via solid state relay,

and thus the power consumption is additive. Maximum current consumption was calculated to be:

Therefore it was concluded that the four heaters could all be powered from one mains plug socket,

protected by a 13 amp fuse.

24

Figure A1.5 - Photograph of the solid state relay control boxes and their connections to IKA

Digital I/O controller.

Figure A1.6 - Photograph of a Crydom D2450 solid state relay. Each relay control box contains

4 D2450 relays.

The 230 V AC power to the 3 x band heaters and 1 x ring heater is provided via Crydom D2450 solid

state relays (Figure A1.6). A copy of the solid state relay specification is presented in Section A3.

One solid state relay was assigned to each heater, which could be configured to form a series of

independent closed loop control systems with temperature feedback. However for the planned

application, the calorimeter was configured to allow all three band heaters to respond to one

temperature feedback input. This is diagrammatically shown in Figure A1.7. The solid state relays

are housed within a ventilated plastic enclosure to isolate the relays from the operator and reduce the

risk of electric shock (Figure A1.5).

Solid state relay

control boxes Interface between the solid state relay control box and the

IKA I/O control.

25

The calorimeter carries out sample temperature tracking by the use of closed loop feedback control.

In adiabatic mode the control loop set point temperature is determined by the temperature of the

sample under test that is resident in the test can. The sample temperature is measured via a K-type

thermocouple which is designed into the test can. Feedback temperature is provided by a single K-

type thermocouple. It should be noted though that the calorimeter has the capability to run four closed

loop temperature control loops, one for each control loop.

The feedback control thermocouple was mounted on the side of the calorimeter body in the gap

between the band heater fasteners. For thermal conduction, the thermocouple junction was encased in

copper plate. A single thermocouple was located on the top plate copper disk which houses the ring

heater. Figure A1.7 shows the control loop method adopted. A larger annotated version of this block

diagram is available in Section A2.

A ‘global’ set point temperature can be manually presented to the calorimeter to instigate a thermal

runaway event or to establish a benchmark auto ignition temperature. The control and data

acquisition system can then be switched to sample temperature tracking as the reaction / auto ignition

event begins.

The Proportional, Integral and Derivative (PID) controller, as used by the heater’s closed loop control

systems, were set up as P-D controllers (proportional-derivative). A limited amount of controller

tuning was conducted, but there is scope to further refine these controllers to achieve critical damping

of the overall system. Currently the PID controllers have the P, I and D settings shown in Table A1.

Figure A1.7 - Block diagram to show control and electrical power flow through the

calorimeter.

26

Table A1 - The P, I and D values used by each control loop.

Controller ID Proportional (P) Integral (I) Derivative (D)

PID 1 1 1000000 10

PID 2 not used not used not used

PID 3 3 1000000 10

PID 4 not used not used not used

The “I” term was eliminated by making it a very large value. In practice this meant that the controller

fell short of the set-point. This was not an issue because the compound heating effects of the

neighbouring band heaters brought the system into the set-point temperature region with less over

shoot than if an “I” term was also used.

The “D” term of 10 was chosen for both of the controllers used, because tests showed that this value

provided an acceptable recovery time after a loop disturbance, which allowed the “P” values to be

increased further without suffering control loop oscillation.

The output from the PID controller is fed into a Pulse Width Modulation (PWM) function. This

software driven function calculates an “on” period in the operating time window, based upon the PID

controller output. The output of the PWM function is a percentage of the operating time window,

referred to as the duty cycle, where mains power is provided to the heaters. Using this function, as the

set-point temperature is approached, the power provided to the heaters is progressively reduced on

every cycle until the set-point temperature is met. The output from the PWM is fed to the solid state

relay via an IKA IO controller, which is a digital control panel providing a 0 V or 5 V signal. The

output of the solid state relay control box is a 230 V AC, under load.

A1.2.1 - Heaters & Hotplate Figure A1.8 show the model of band heater used and the specification, as stated by the supplier,

Omega Engineering UK. Each heater has an independent 230 V AC mains supply, provided via a

Crydom D2450 solid state relay.

Figure A1.8 - Photograph to show the type of band heater clamped around the calorimeter

body.

Specification of the heater shown in Figure A1.8:

Omega Engineering Part Number: DB-050772

240 V One-piece band heater, barrel diameter 5”, 770 W

Heater 1

Heater 2

27

The calorimeter has the capacity to use four band heaters. Currently only three band heaters are used.

Each band heater is made up of two heater elements, joined at one end by a semi-malleable metal. At

the other end a bolt fastener allows the band heater to be tightly secured around the calorimeter body.

Each side of the heater can be powered separately, but in this application, the two heaters have been

wired in parallel to allow them to be powered from a single solid state relay. Figure A1.9 shows the

wiring of a band heater.

Figure A1.9 - Photograph to show the band heaters wired to the solid state relay.

Figure A1.10 - Photograph to show the type of disk heater used in the top plate assembly.

Specification of the heater shown in Figure A1.10:

Omega Engineering Part Number: A-205/240

A Series Ring Heater 240v, 500w Single-Heat Element Chrome Steel Sheath

The calorimeter uses one Ring Heater, which is part of the top plate assembly. This provides the

required heating for the top of the calorimeter chamber.

Heater 2

terminals

Cables to solid

state relay

Heater 1

terminals

28

Figure A1.11 - Photograph to show the disk heater in-situ in the top plate assembly.

A1.2.2 - Stirrer hotplate The stirrer hotplate purchased from Sigma Aldrich has the description: “IKA RCT basic IKAMAG

safety control. Universal hot plate magnetic stirrer, 20 L, 1500 rpm, 230 v, 1c/s”.

The hotplate stirrer is controlled by the Labworldsoft control and logger software. The controls as

shown in Figure A1.12 below are inactive, and any change made using them is overridden by the

Labworldsoft program.

Figure A1.12 - Photograph to show the front control panel of the hotplate stirrer and its position

in the calorimeter assembly.

Ring Heater, in top

plate assembly

Cables to solid

state relay

29

Figure A1.13 shows the connection between the hotplate stirrer and the Labworldsoft program. The

figure also highlights the PT100 resistance thermometer used in conjunction with the hotplate stirrer,

which indicates the temperature in the lower area of the calorimeter chamber. It should be noted that

by design the hotplate stirrer is limited to 300°C.

Figure A1.13 - Photograph to show the PT100 resistance thermometer and Labworldsoft

control cable connection to the hotplate stirrer.

The hotplate temperature is monitored using a PT100 (resistance thermometer), as shown in Figure

A1.14. The PT100 probe is housed in a push fit hole which goes through to the centre of the

calorimeter bottom plate; it is in direct contact with the IKA hotplate stirrer plate surface.

Figure A1.14 - Photograph to show the position of the PT100 resistance thermometer in-situ in

the calorimeter hotplate.

PT100 temperature

measurement device

Communication to

Labworldsoft

PT100

measurement

position

30

A1.3 - Temperature Measurement Figure A1.15 shows the temperature transmitters located adjacent to the calorimeter. Temperature is

monitored at five locations inside the instrument. The sixth transmitter, as shown in Figure A1.15,

will cater for the closed loop control system for a bottom heater, if required. These are necessary to

monitor the temperatures in different areas of the calorimeter body so that the control software can

extend or reduce the “on” period of the heaters when tracking test sample temperatures. The

temperature measurements are part of the closed loop control system (see Section A2 - Calorimeter

Control System).

Figure A1.15 - Photograph to show the temperature transmitters used in conjunction with the

Type-K control thermocouples.

All of the temperature transmitters used by the calorimeter are ranged to operate optimally in the 0 –

500oC temperature region, and they all only accept K-type thermocouples.

The measured temperatures are transmitted back to the Labworldsoft control and logger software via

the IKA DC2 I/O modules (Figure A1.16). The 4-20 mA currents, as measured by the temperature

transmitters, are appropriately scaled by Labworldsoft before being recorded as actual temperature

measurements. The temperature transmitters all have in-built cold junction compensation.

31

Figure A1.16 - Photograph to show connection from the Labworldsoft PWM function, via the

Datacontrol I/O2 to the switch signal input terminals of the solid state relays, as show in Figure

A1.6.

At the clamp point of each heater, a thermocouple plate was secured tightly against the calorimeter

body. This allowed the control temperature in the respective calorimeter heating band zone to be

measured. In addition to the three band heaters around the calorimeter body, thermocouples were also

placed to measure the temperatures of the top plate, bottom plate and test sample (Figure A1.17).

Figure A1.17 - Photograph to show the positioning on the calorimeter body of the Type-K

thermocouples that are used in the feedback control system.

The thermocouples are held in positioned in a push tight hole in a separate copper plate that sits in the

fastening gap. This ensures good thermal conduction between the calorimeter body and the

thermocouple, therefore increasing the accuracy of the overall temperature measurement. The copper

plate was secured into place, using four bolts, through a threaded steel plate which applies force to the

copper plate, bending it to the contour of the calorimeter body (Figure A1.18). It is important that the

32

copper plate sits tightly against the calorimeter body and air gaps are not present, which can lead to

inaccurate temperature measurement.

Figure A1.18 - Diagram to show cross-section from above of the securing mechanism of the

thermocouples used to measure calorimeter body temperature. The hole represents the location

of the thermocouple.

A1.4 - Pressure Transducer Changes in test can pressure can be measured when both the inlet and outlet valves are closed. This

calorimeter measures pressure using a 0 – 5 Bar (Abs) transducer (Figure A1.19). The transducer is a

4-20 mA device and connects to Labworldsoft via a fuse protected 250 resistor circuit contained in

the instrumentation cabinet. The IKA DC2 data acquisition module measures the voltage between the

resistor and ground. The voltage is then ranged between 0-5 V before being logged.

Figure A1.19 - Photograph to show the position of the pressure transducer on the calorimeter.

The pressure transducer is positioned on the inlet side of the test can via a Swagelok tee piece.

Copper plate

Thermocouple hole Calorimeter Body

Steel plate

33

A1.5 - Calorimeter test chamber assembly The body of the calorimeter is a single piece of copper pipe which has been machined to width,

reducing the calorimeter’s heat capacity, thereby improving the heating response time. Figure A1.20

shows the dimensions (mm) of the copper pipe that was used for the test chamber.

Figure A1.20 - Diagram to show the overall dimensions of the copper pipe which was used to

form the calorimeter body.

Figure A1.21 shows the actual calorimeter copper pipe with the top plate assembly removed during

test preparation.

Figure A1.21 - Photograph to show the calorimeter inner cavity from above. This view is from

where the top plate, top heater and sample reservoir would normally be sited.

200

120

140

mm

160

34

A1.6 - Top Plate assembly The top plate assembly consists of:

1. A ring heater

2. A copper disk, with diameter 115 mm

3. Top plate thermocouple

4. Phi-TEC calorimetric test can

5. Swagelok connections on the inlet and outlets

6. Swagelok tee piece on the inlet line for connection of pressure transducer.

Figure A1.22 - Photograph to show the top plate, top heater and sample test can assembly.

The test sample is injected through a valve and tube feed through the insulated cavity and finally into

the test can. The test can contains an integrated type k thermocouple (Figure A1.22). The cans used

are the same as those used by the HEL Phi-TEC calorimeter1. The assembly, shown in Figure A1.22,

is lowered into the calorimeter test cavity shown in Figure A1.21.

Figure A1.23 shows the sample reservoir in which the oil under test is held immediately prior to being

injected into the test can. The sample reservoir is seated on a Swagelok ferrule which allows the

vacuum pump or compressed air line to be connected allowing the test can to be cleaned.

1 http://www.helgroup.com/reactor-systems/thermal-hazards-and-calorimetry/phitec-ii/. Viewed on 30/05/2013.

1.

2.

3.

4.

5.

6.

35

Figure A1.23 - Photograph to show the sample reservoir attached to the inlet pipe of the test

can.

Figure A1.24 shows the piped connections to the ‘hastelloy’ sample can for sample injection,

evacuation and flushing using compressed air. This is the same test can as used by the HEL Phi-TEC

calorimeter. In order to position the test can in the top plate assembly, the thermocouple mini

connector has to be removed so that the sheath can be passed through the top plate hole. Once in

place, the appropriate Swagelok fitting and ferrules are slid onto the thermocouple, before the mini

connector is re-attached. This process has to be repeated every time a test can is replaced.

Figure A1.24 - Photograph showing an example of the test can used in the calorimeter.

The inlet, outlet and thermocouple are pre-attached to the test can. Internally a stirrer bar is present

that be agitated by the hotplate stirrer.

36

A1.7 - Control & data logging software The calorimeter is controlled using a Labworldsoft program which interfaces using two IKA DC2

data acquisition modules and an IKA I/O controller. Figure A1.25 shows the flow of data between the

calorimeter hardware and the Labworldsoft program.

SSR Box 11Top plate heater

Band Heater 1

Band Heater 2

Band Heater 3

IKA Hotplate & Stirrer

PT100

SSR Box 12

SSR Box 13

SSR Box 14

LabworldsoftControl

Program

Drawn: R Braddock HSL 10/12/2012Version: 1.01

lilac

blue

white

black

l.greenbrown

pink

grey

Control I/O2Chan 4

Control I/O2Chan 3

Control I/O2Chan 2

Control I/O2Chan 1

IKA Digital I/O controllerP

Transmitter 1

Transmitter 2

Temperature TransmittersType-K (0-500 degC)

Transmitter 3

Transmitter 4

Transmitter 5

lilacblue

whiteblack

green

pinkgrey

orangered

yellow

l.green

brown

Test Cancf35

37

cf33

35

cf41

43

cf40

42

cf39

41

cf38

40

DC2-2Chan 3

DC2-1Chan 4

DC2-1Chan 3

DC2-1Chan 2

DC2-1Chan 1

DC2-2Chan 4

Figure A1.25 - Block diagram to show the interfacing between the hardware and software of the

calorimeter control system.

It should be noted that Figure A1.25 shows the system configured to use all four closed loop

controllers. In reality Band Heaters 1-3 and the top plate heater are all controlled by the PWM signal

provided by Control I/O2, channel 3.

Figure A1.26 shows the Labworldsoft program that controls the calorimeter hardware. The main

features of the program are:

Heat ramping - This feature is used to bring the calorimeter hardware up to operating temperature in

a controlled manner, minimising set-point temperature overshoot. This feature can be tailored for use

with adiabatic measurement so that the calorimeter is brought up to a temperature which is

predetermined to be under the auto ignition temperature of the substance under test.

Temperature tracking (adiabatic measurement) - In this mode of operation the measured

temperature of the substance under test is “followed” by the calorimeter’s heaters.

Calorimeter Hardware Control interface & software

PWM Control Signal

Temperature feedback

37

Temperature set-point (isothermal measurement) - In this mode of operation, a predetermined set

point is entered into the program. Using PWM, the heaters warm the calorimeter to this set-point

temperature. The accuracy to which the set point temperature is met is dictated primarily by the tuning

of the PID control loop filters. The PID loops are tuned as stated in Table A1.

The software uses the following functions in order to achieve the measurement techniques:

PID controller - The larger the proportionality factor P, the faster the system deviation settles.

However P should not be too large to prevent control loop oscillation. Any remaining control loop

oscillation can be reduced further by the integral portion I. The controller dynamic is significantly

improved by the differential proportion D. The proportionality factor P may be greater than for a pure

P or PI controller without the regulator circuit becoming unstable.

PWM function - The Pulse Width Modulation (PWM) function calculates an “on” period in the

operating time window, based upon the PID controller output. The output of the PWM function is a

percentage of the operating time window, referred to as the duty cycle in which power is provided to

the designated digital output channels.

Figure A1.26 - Screenshot to show the Labworldsoft program function module interconnection.

All temperatures and duty times are displayed on the screen for the user to see and respond to (Figure

A1.27). The data are also written to a comma separated text file which is time and date stamped and

is readily processed using Microsoft Excel or similar spread sheet or data processing software.

38

Figure A1.27 - Screenshot to show the Labworldsoft user interface.

On the Labworldsoft display, each heater control loop has an associated temperature, control voltage

and duty cycle indicator (Figure A1.27). There are two primary controls on the display, the “T set”

slider which allows the user to specify a set point temperature, and the “RPM set” slider which allows

the user to specify a required hotplate stirrer revolution requirement. All measured temperatures are

viewable in real time via the temperature versus time graph, labelled “y/t graph” in Figure A1.27.

39

A2 - CALORIMETER CONTROL SYSTEM

40

Heater

Equipment under

Test (EUT)

PID

Control

PWM (v)

0-100%

0-240v

(AC)

EUT follower (DegC)

Heater

Chamber wall Temperature 1 (DegC)

Mains voltage (240v AC)

Low voltage (0-24v DC)

Thermocouple (Type-K)

Set-point

(Spt1)

Isothermal Testing

Adiabatic Testing

PWM (v)

0-100%

PID

Control Disk Heater

Top disk Temperature 2 (DegC)

Figure A2.1 - Block diagram to show the control system principle to be used on the auto ignition calorimeter. V1.00

41

Heater

Equipment under

Test (EUT)

Set-point

(Spt1)

PID

Control

PWM (v)

0-100%

0-240v

(AC)

EUT follower (DegC)

Heater

Internal Temperature 1 (DegC)

PID

Control

PWM (v)

0-100%

0-240v

(AC)

Internal Temperature 2 (DegC)

… … … … Mains voltage (240v AC)

Low voltage (0-24v DC)

Thermocouple (Type-K)

Chamber Temperature 1 (DegC)

Alternative SP feeds

Figure A2.2 - Block diagram to show the control system principle to be used on the auto ignition calorimeter - Original independent control.

V1.01

42

Heater

Heater

Heater

Heater

Band heater max output 770W

Band heater max output 770 W

Band heater max output 770 W

Band heater max output 500W

Calculation of maximum power consumption

Power = Voltage x Current

W = V x A

43

A3 - CRYDON D2450 SOLID STATE RELAY SPECIFICATION DATA SHEET

Series 1 240 VAC • Crydom's signature family of solid-state relays. Ratings from 10A to 125A @ 24-280 VAC

• SCR output for heavy industrial loads

• AC or DC control

• Zero-crossing (resistive loads) or random-fire (inductive loads) output

PRODUCT SELECTION Control Voltage 10A 25A 50A 75A 90A 110A 125A 3-32 VDC D2410 D2425 D2450 D2475 D2490 D21110 D24125

90-280 VAC A2410 A2425 A2450 A2475 A2490 A24110 A24125

18-36 VAC A2410E A2425E A2450E A2475E A2490E A24110E A24125E

OUTPUT SPECIFICATIONS (1) Description 10A 25A 50A 75A 90A 110A 125A Operating Voltage (47-63Hz) [Vrms] 24-280 24-280 24-280 24-280 24-280 24-280 24-280

Transient Overvoltage [Vpk] 600 600 600 600 600 600 600

Maximum Off-State Leakage Current @ Rated Voltage [mArms] 10 10 10 10 10 10 10

Minimum Off-State dv/dt @ Maximum Rated Voltage [V/μsec] (2) 500 500 500 500 500 500 500

Maximum Load Current (3) [Arms] 10 25 50 75 90 110 125

Minimum Load Current [mArms] 40 40 40 40 40 150 150

Maximum Surge Current (16.6ms) [Apk] 120 250 625 1000 1200 1500 1750

Maximum On-State Voltage Drop @ Rated Current [Vpk] 1.6 1.6 1.6 1.6 1.6 1.7 1.7

Thermal Resistance Junction to Case (Rjc) [°C/W] 1.48 1.02 0.63 0.31 0.28 0.25 0.22

Maximum I2 t for Fusing (8.3 msec) [A2 sec] 60 260 1620 4150 6000 9340 12700

Minimum Power Factor (with Maximum Load) 0.5 0.5 0.5 0.5 0.5 0.5 0.5

INPUT SPECIFICATIONS (1) Description ( D PREFIX) ( A PREFIX) ( E SUFFIX) Control Voltage Range 3-32 VDC 90-280 Vrms 18-36 Vrms

Maximum Reverse Voltage -32 - -

Maximum Turn-On Voltage 3.0 VDC 90 Vrms 18 Vrms

Minimum Turn-Off Voltage 1.0 VDC 10 Vrms 4.0 Vrms

Typical Input Current 3.4-20 mA 2.0-4.0 mA 3 mA

Nominal Input Impedance [Ohms] 1500 Ohm 60 K Ohm 9.0 K Ohm

Maximum Turn-On Time [msec] (4) 1/2 Cycle 10 10

Maximum Turn-Off Time [msec] 1/2 Cycle 40 40

GENERAL SPECIFICATIONS Description Parameters Dielectric Strength, Input/Output/Base (50/60Hz) 4000 Vrms

Minimum Insulation Resistance (@ 500 V DC) 109 Ohm

Maximum Capacitance, Input/Output 8 pF

Ambient Operating Temperature Range -40 to 80 °C

Ambient Storage Temperature Range -40 to 125 °C

Weight (typical) 3 oz (86.5g)

Encapsulation Thermally conductive Epoxy

Terminals Screw and saddle Clamps Furnished, Unmounted

Recommended Terminal Screw Torque Range: 6-32 Screws - 10 in/lbs. 8-32 & 10-32 Screws - 20 in. lbs. (Screws dry without grease)

Fastons: Single pair (up to 25A) Double pair* (50A model only) *Caution: User must connect to both pairs

GENERAL NOTES 1) All parameters at 25°C unless otherwise specified.

2) Off-State dv/dt test method per EIA/NARM standard RS-443, paragraph 13.11.1

3) Heat sinking required, for derating curves see page 3.

4) Turn-on time for Random turn-on versions is 0.02 msec (DC Control Models)

AGENCY APPROVALS UL E116949

CSA LR81689

VDE 10143 UG (Not Applicable: -B and 4D) Rev. 030609

44

Published by the Health & Safety Executive 12/16

Spontaneous ignition of gas turbine lubricants at temperatures below their standard auto-ignition temperatures

RR1076

www.hse.gov.uk

There have been a number of incidents resulting in lubricating oil leaking in offshore gas turbine enclosures which could ignite if they came into contact with hot surfaces below their Auto Ignition Temperature (AIT). To assess the risk of auto-ignition, standard minimum AITs are used. However, AITs under industrial conditions are difficult to calculate and can be less than these standard values.

This report describes research using a Spontaneous Combustion Calorimeter developed to study spontaneous ignition. Preliminary tests were done for a range of process conditions that can influence minimum AITs for a number of gas turbine lubricating oils. These showed that ignition can occur at temperatures well below the standard minimum AIT. This indicates that if manufacturers rely on standard AITs at the design stage of gas turbines and enclosures, it may lead to a system that is likely to increase the ignition probability of any flammable release. To confirm these findings, further tests would be needed over a wider temperature range and under conditions which more closely represent the conditions in gas turbine enclosures.

Until AITs under industrial conditions are understood and addressed in design criteria, dutyholders will need to err on the side of caution in identifying and adequately controlling potential ignition sources.

This report and the work it describes were funded by the Health and Safety Executive (HSE). Its contents, including any opinions and/or conclusions expressed, are those of the authors alone and do not necessarily reflect HSE policy.